This forum is disabled, please visit https://forum.opencv.org
 | 1 | initial version |
I took the liberty of bringing up this really old question as I seem to have a similar problem, but I do not understand the accepted answer. Should I normalize the image points to [-1,1]? It is not helping at all. If I provide native screen coordinates, EPNP works perfectly for a perfect case, however, when I add even 1 pixel error to just one of the 4 points, the results are crazy. I checked the Matlab version of EPNP and the results are fine.
